Impairment Navigation through Systematic Investigation & Glance Habit Tracking

Purpose and goal

INSIGHT aims to improve traffic safety through a system that detects impaired driving in real-time using data analysis of multi-modal data. The goal is a predictive model that, based on real driving data, such as gaze behavior, identifies deviations and warns the driver.The project supports safer vehicle technologies, policies, EuroNCAP, and EU regulations.

Expected effects and result

The project strives to provide measurable results with the target of less than 10 false impairment events per 1 million km and providing 95% correct detection of drowsiness in the combine dataset consisting of EuroFOT, NTHU, and SleepEye (Vinnova) datasets. As well the project targets correct detection of 95% of impairment event in the newly collected INSIGHT impairment dataset that includes mimicking impaired behaviour like severe distraction and medical problems target.

Planned approach and implementation

WP1 ensures project management and coordination, including reporting and steering meetings. WP2 focuses on literature studies on impairment detection. WP3 handles data preparation, and exploration of real and simulated datasets. WP4 develops models of normal driving behavior. WP5 creates algorithms for impairment detection using multimodal analysis and Explainable AI. WP6 validates and refines the system through controlled testing and vehicle integration.